Vol. 465: 299–304, 2012 MARINE ECOLOGY PROGRESS SERIES Published September 28 doi: 10.3354/meps09902 Mar Ecol Prog Ser OPENPEN ACCESSCCESS AS WE SEE IT Fish ingredients in online recipes do not promote the sustainable use of vulnerable taxa Charis Apostolidis*, Konstantinos I. Stergiou Laboratory of Ichthyology, School of Biology, Aristotle University of Thessaloniki, Thessaloniki 54124, Greece ABSTRACT: Consumer preferences can be influenced by cultural factors, such as recipes. Thus, popular recipes using seafood ingredients could encourage demand for the featured marine spe- cies. We recorded the fish taxa appearing in all online Greek recipes and related the number of recipes in which a taxon is a suggested ingredient to the taxon’s landings and vulnerability to fish- ing. For marine fish, the number of recipes per taxon (henceforth called recipe prevalence) increased significantly with the taxon’s landings. When the effect of landings was removed, recipe prevalence increased significantly with vulnerability to fishing, indicating that more vulnerable taxa appear more frequently in recipes. This pattern should be reversed by eliminating the recipe abundance of taxa that are very vulnerable to fishing and counterbalance such a reduction with an increase in the representation of taxa of low vulnerability, landed in large quantities and cur- rently underrepresented in recipes. Managing recipe composition by educating a small group of stakeholders (i.e. chefs and recipe authors) may be a cost-effective way of promoting fisheries sustainability when compared to directly educating consumers. KEY WORDS: Consumer influence · Conservation · Landings · Recipes · Vulnerability to fishing Resale or republication not permitted without written consent of the publisher INTRODUCTION management and aim to influence consumer pre - ferences towards sustainability (Halweil 2006, Most fisheries are depleted and/or are being con- Jacquet & Pauly 2007, Roheim 2009). However, the tinually overfished (Garcia & Rosenberg 2010). Irre- effect of cultural factors, such as recipes, on consumer spective of the magnitude of fisheries mismanage- choices regarding fish consumption has been largely ment (Myers & Worm 2003, Pauly et al. 2003, Hilborn ignored (Levin & Dufault 2010). 2007, Worm et al. 2009), excessive fishing impairs Nowadays, chefs have started taking an energetic ecosystem health (e.g. Jennings & Kaiser 1998, Pauly role in the conservation of fish stocks by recommend- et al. 1998, Stergiou 2002) and has important socio- ing the consumption of sustainably exploited fish, economic implications (e.g. Failler & Pan 2007, Srini- with positive results in most cases. For example, after vasan et al. 2012). Thus, there is a growing concern the launch of the BBC Channel 4 show ‘Fish season’, for the future of fisheries, which, among other actions, fish demand considerably increased, especially for has resulted in the launch of various consumer and species that were proposed to replace the overfished market-based actions, such as public information Atlantic cod Gadus morhua (e.g. demand for pollock campaigns and sustainability certification schemes Pollachius virens was reported to increase by 167%) (Halweil 2006, Jacquet & Pauly 2007, Roheim 2009). (Tyler 2011). However, such actions have not yet These campaigns complement traditional fisheries been broadly applied to recipe websites. The impor- *Email: [email protected] © Inter-Research 2012 · www.int-res.com 300 Mar Ecol Prog Ser 465: 299–304, 2012 tance of online recipes in influencing the public can expert system (see Cheung et al. 2005 for a detailed be illustrated by the fact that the most-visited cook- description). The estimation of vulnerability is based ing website, allrecipes.com, attains about 24 million on the species’ life history (i.e. maximum length, age unique visitors monthly. Online recipes are not only a at first maturity, longevity, von Bertalanffy growth source of public influence but, to a certain extent, parameter K, natural mortality rate and fecundity) could also reflect consumer preferences, since users and ecology (i.e. strength of spatial behavior and can post their own recipes. geographic range). Vulnerability is a metric of Recipes have not been evaluated on a large scale extinction risk and ranges from 0 (lowest) to 100 from a sustainability point of view. Testing the rela- (highest). Mean landings per taxon for 1990 to 2007 tion between the prevalence of different fish taxa in were estimated from the landings reported by recipes and their availability (landings) and vulnera- Moutopoulos & Stergiou (2012). All scientific names bility can reveal whether recipes could act as a driver were obtained from FishBase (Froese & Pauly 2012). for unsustainable use of marine resources. In addi- The relation between recipe prevalence, landings tion, the form of the relation could provide directions and vulnerability was described using stepwise for the conservation of fish stocks. Here, we tested multiple regression, and the reported p-values are this relation using the marine fish taxa in all online 2-tailed. Greek recipes. RESULTS MATERIALS AND METHODS Of the 3858 recipes found, 179 (4.6%) referred to We analyzed the prevalence of different fish taxa fish without specifying a taxon and thus were in Greek recipes and related the number of recipes in excluded from our analysis. In the remaining 3679 which each taxon appears (henceforth called recipe recipes, there were 4265 occurrences of identified prevalence) with the taxon’s landings and intrinsic taxa (i.e. 1.1 fish per recipe, as some recipes sug- vulnerability to fishing. We examined all online gested >1 taxon) corresponding to 80 Greek common recipes that were available on every website listed in names, 2 of which (perch and tilapia) could not be the catalogs of 2 Greek cooking recipe search assigned to a specific taxon. This is because these 2 engines (www.sintagoulis.gr and www.cookle.gr, taxa, both derived from aquaculture, are sold without accessed December 2011). any information on scientific names. The remaining Overall, we collected 3858 recipes using fish as a 78 common names were assigned to 71 taxa, of which primary or secondary ingredient. These recipes were 60 (84.5%) were at the species level, 8 (11.3%) at the derived from 129 websites, covering a wide variety of genus level and 3 (4.2%) at the family level (Table sources (i.e. recipe web sites, cooking blogs, news - A1). Overall, 15 taxa (20.55%) accounted for 3234 papers, online magazines and television cooking pro- (75.8%) of the occurrences (Fig. 1). grams). In the vast majority of cases, no information was provided regarding the date of recipes or whether recipes posted on the web were actually taken from other sources (e.g. from a published recipe book). All fish common names were assigned to the lowest possible taxon. European hake Merluccius merluccius and Gadus morhua have the same Greek common name (‘bakaliaros’). The former is only sold fresh, while the latter is sold as salted fillets or frozen. Thus, ingredients referring to fresh ‘bakaliaros’ were assigned to M. merluc- cius and the remaining ones to G. morhua. Intrinsic vulnerability to fishing, hence- forth called vulnerability, was taken from Fig. 1. Number of recipes (bars) and cumulative percentage (line, right FishBase (www.fishbase.org; Froese & y-axis) by rank for all fish taxa (n = 71) cited as ingredients in Pauly 2012), as estimated by a fuzzy logic Greek online recipes Apostolidis & Stergiou: Recipes and conservation of fish stocks 301 Fig. 2. Distribution of (a) recipe prevalence (number of recipes in which each taxon appears), (b) vulnerability and (c) mean landings for the fish taxa cited in Greek online recipes. x-axis labels denote midpoints of bins The distributions of recipe prevalence, vulnerabil- ability and mean landings was stronger (R2 = 0.405, p ity and mean landings for the different taxa are < 0.001, df = 2,31, F = 10.529): log(recipe prevalence) shown in Fig. 2. Recipe prevalence ranged from 1 = −3.223 + 0.444 log(mean landings) + 1.940 log(vul- (0.02%) to 595 (14.0%) recipes (Fig. 2a, Table A1), nerability) after excluding 3 apparent outlier taxa with a mean value (±SD) of 58 ± 105.33. Vulnerability (Engraulis encrasicolus, Solea solea and white ranged from 25, for European anchovy Engraulis grouper Epinephelus aeneus) above the regression encrasicolus, to 88, for sturgeon Acipenser sturio and line and 4 taxa (blotched picarel Spicara maena, Atlantic halibut Hippoglossus hippoglossus, with a annular seabream Diplodus annularis, blue whiting mean value of 52.8 ± 16.52 and a main mode at 40 to Micro mesistius poutassou and round sardinella Sar- 45 (Fig. 2b). Mean landings (Fig. 2c) ranged from dinella aurita) below the regression line. 84.3 t, for shi drum Umbrina cirrosa, to 17798 t, for European sardine Sardina pilchardus. We confined our analysis to marine wild stocks DISCUSSION native to Greece (57.8% of recipe citations), exclud- ing cultured (25.6%), imported (15.4%) and freshwa- The fact that recipe prevalence increased with ter species (1.2%). The most prevalent of the 55 mar- landings indicates that the regional cooking culture ine taxa cited in 2467 recipes were Engraulis is developed based on the more abundant species in encrasicolus, followed by Thunnus spp. and common the area. Yet, recipe composition applies a greater sole Solea solea, with 286 (11.6%), 284 (11.5%) and 226 (9.2%) occur- rences respectively (Table A1). For the 41 Greek native marine taxa for which landings were avail- able, the following multiple regres- sion was found (R2 = 0.216, p = 0.039, df = 2,38, F = 5.230): log(recipe prevalence) = −2.852 + 0.471 log (mean landings) + 1.645 log(vulner- ability). The 2 predictors were not intercorrelated (n = 41, r = 0.176, p = 0.271). Thus, after correcting for the effect of landings (partial r = 0.409, p = 0.009), recipe prevalence was significantly correlated (partial r = 0.322, n = 41, p = 0.040) with vulner- ability, indicating that more vulner- able species appear more frequently Fig.